© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Qwen3-4B
    5. 20-TRANSCODER-HP
    7. 140938
    Prev
    Next
    INDEX
    Explanations
    No Explanations Found
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_20.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.20.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    icolor
    -0.27
    é¹Ĭ
    -0.26
    .shiro
    -0.25
     prelim
    -0.25
    auer
    -0.25
     hopefully
    -0.25
    .baidu
    -0.24
     Los
    -0.24
    rosse
    -0.24
     arresting
    -0.24
    POSITIVE LOGITS
    ista
    0.26
    OSP
    0.25
     santa
    0.25
    ç«ĭ
    0.25
     metres
    0.25
    å¡ŀ
    0.25
    ISTA
    0.25
    isme
    0.24
    çİ°åľº
    0.24
     Sass
    0.24
    Activations Density 5.056%

    No Known Activations

    This feature has no known activations.